What do older people think that others think of them, and does it matter? The role of meta-perceptions and social norms in the prediction of perceived age discrimination.

Meta-Perceptions

Meta-perceptions constitute a type of knowledge that is thought to be widely shared by others ( Vorauer, Main, & O’Connell, 1998 ), and can be distinguished at the individual- and societal-levels. In this article, individual-level or personal meta-perceptions describe older people’s personal beliefs about how others in society perceive their age group. Research on meta-cognition and intergroup relations shows that individuals’ negative meta-perceptions may elicit heightened perceptions of discrimination ( Frey & Tropp, 2006 ; Vorauer et al., 1998 ). Negative meta-perceptions inhibit intergroup contact by increasing anxiety about interactions with outgroup members (e.g., in simulated contact situations; Finchilescu, 2010 ) and increase potential for miscommunication and tension ( Vorauer & Sakamoto, 2008 ). Individuals may infer from meta-perceptions whether they are liked by the outgroup and are likely to be targets of discrimination. Negative meta-perceptions are especially likely to be held by members of stigmatized groups ( Pinel, 1999 ; Richeson & Shelton, 2006 ; Vorauer et al., 1998 ) as seems likely for older people. It is known that members of some stigmatized groups expect to be seen negatively by others and are more likely to perceive events as discriminatory ( Pinel, 1999 ). This link has not been empirically tested yet or explored in relation to age despite the clear theoretical and empirical implication that older people’s personal meta-perceptions should be related to perceptions of age discrimination (PAD; Cuddy, Fiske, & Glick, 2007 ; Fiske, Cuddy, Glick, & Xu, 2002 ; Garstka et al., 2004 ).

Meta-perceptions are often assessed by asking respondents how a specific social group is viewed by most people in society (see, e.g., Fiske et al., 2002 ). Strictly speaking, these measures tap individual differences in meta-perceptions from which researchers may infer other individual differences in attitudes or behavior (cf. Chiu, Gelfand, Yamagishi, Shteynberg, & Wan, 2010 ). However, a higher order feature of meta-perceptions is the actual shared cultural perception which is represented by aggregating all individuals’ perceptions within the pertinent cultural context ( Schwartz, 2013 ). These societal meta-perceptions represent societal views, reflecting the extent to which the majority in a cultural community shares a particular perception about a social group.

Empirical analyses on societal meta-perceptions have been limited to the study of media content (e.g., Donlon, Ashman, & Levy, 2005 ) or to small samples within a single country (for a review, see Kite, Stockdale, Whitley, & Johnson, 2005 ) or to cross-national student samples ( Löckenhoff et al., 2009 ). Hence, the present research is the first to examine personal and societal meta-perceptions in large scale representative samples across multiple countries. We draw on the multilevel analysis literature (e.g., Bliese & Jex, 2002 ) to understand how personal and societal meta-perceptions may predict PAD.

In this article societal meta-perceptions represent the societal aging climate and are generated by aggregating individual (out) group members’ responses and calculating the group means. Our study examines cross-level direct-effect models ( Bliese & Jex, 2002 ), that is, the link between meta-perceptions at the individual- and societal-level and PAD as an individual-level outcome variable. To do so, we will simultaneously regress older people’s PAD on both personal and societal meta-perceptions. If societal meta-perceptions are significantly related to PAD even when older people’s personal meta-perceptions are also in the model, this will demonstrate that there are distinguishable implications of the two levels of meta-perceptions. It is important to note that it would support the idea that societal meta-perceptions have emergent properties.

Theoretical Considerations

The benefit of simultaneously modeling direct effects of individual- and societal-level constructs lies in the integration of different theoretical conceptualizations. The social representation perspective ( Deaux & Philogène, 2001 ; Moscovici, 1988 ) suggests that societal level meta-perceptions influence PAD because, for example, if there is a social consensus that older people are incompetent, then older people should perceive age discrimination. Hence, it is an objective conceptualization of aging beliefs in which a certain societal aging climate is seen as universally negative for individuals’ psychological aging experience. Alternately, the meta-cognition perspective (e.g., Frey & Tropp, 2006 ; Vorauer et al., 1998 ) is a subjective conceptualization of aging beliefs by suggesting that negative meta-perceptions about aging relate to PAD, but only if they exist as personal meta-perceptions. Although both perspectives suggest that negative meta-perceptions about aging should be linked to PAD, they address different conceptualizations, and therefore, different levels of analyses. Thus far they have been considered separate in research; whereas in this article we consider that they both may have an influence in explaining PAD.

When meta-perceptions are examined at the individual- and societal-level, there is the possibility that they can produce congruent or distinctive influences on PAD. This is because the two levels of analyses are statistically independent ( Bliese & Jex, 2002 ; see also Fischer, Vauclair, Fontaine, & Schwartz, 2010 ) and, therefore, the individual- and group-levels of meta-perceptions can exert independent effects on the criterion variable. The extent to which individual-level and societal-level meta-perceptions exert similar effects is known as isomorphism. We believe that there are good reasons to expect isomorphism to occur given that social representations of stereotypic beliefs ( Deaux & Philogène, 2001 ; Moscovici, 1988 ) can have the same impact as personally endorsed stereotypic beliefs. There are two ways in which social consensus about aging beliefs may increase an older person’s vulnerability to age discrimination. First, older people may inadvertently display stereotype-confirming behavior, which can then lead others to behave toward them in ways that further reinforce the stereotype. Second, if a false consensus exists (, e.g., regarding older people’s competence), this is likely to be expressed through conversation, media imagery and in people’s assumptions. It therefore represents an unavoidable ‘threat in the air’ for older people, as posited by stereotype threat theory ( Steele, 1997 ). Stereotype threat can also result in stereotype-confirming behavior such as substantial decrements in cognitive and physical performance ( Lamont et al., 2015 ). This threat then creates a lens through which older people may perceive that they are not accorded the respect and treatment they deserve. As such, societal and personal meta-perceptions should exhibit very similar, or isomorphic, associations with PAD.

For our purposes, we adopt a tripartite model to conceptualize the content of meta-perceptions and age discrimination as part of the ageism syndrome, delineating different types of aging perceptions that have been flagged requiring closer attention in the literature (see Levy, 2009 ). Ageism can be defined as negative or positive stereotypes, prejudice, and/or discrimination against (or to the advantage of) elderly people on the basis of their chronological age or on the basis of a perception of them as being old ( Iversen, Larsen, & Solem, 2009 ). On the basis of social psychological theories, we propose specific hypotheses about how different elements of aging meta-perceptions, tapping into stereotyping and prejudice, should relate to perceptions of discrimination, see Table 1 . 1

Compared with middle-aged adults, older adults are commonly perceived as having lower social status in the sense of power, wealth, respect, influence, and prestige ( Garstka et al., 2004 ). These perceptions affect the way older people are treated in social interactions ( Fiske et al., 2002 ), and older adults are likely to attribute discriminatory practices to their membership in a low status age group ( Garstka et al., 2004 ). Hence, older individuals who believe that their age group is seen as of low social status should also be more sensitive to discriminatory treatment because of age. Moreover, older people are attributed a much higher social status in some countries than in others ( Vauclair et al., 2015b ), which should have an effect on how older people feel they are treated in these societies. Hypothesis 1 (H1): Meta-perceptions of the perceived social status of older people should predict older adult’s PAD negatively at the individual-level (H1a) and societal-level (H1b).

Age stereotyping

The stereotype content model suggests that two underlying dimensions, competence and warmth, organize the stereotypical perceptions associated with any social group in society (see Fiske et al., 2002 ). Ascribed competence is the degree to which a social group is characterized as intelligent and capable and ascribed warmth is the degree to which the social group is regarded as friendly and likable. A negative evaluation of a social group along these two dimensions should be related to discriminatory acts against this group ( Cuddy et al., 2007 ) and therefore to the experience of discrimination. North and Fiske’s (2015) meta-analysis indicated that age stereotypes of competence and warmth evaluations do differ depending on the cultural context.

Hence, we expected the following: Hypothesis 2 (H2): Meta-perceptions of warmth predict PAD negatively at the individual-level (H2a) and societal-level (H2b).

Hypothesis 3 (H3): Meta-perceptions of competence predict PAD negatively at the individual-level (H3a) and societal-level (H3b). 2

Paternalistic prejudice

Previous research has shown that older people are usually pitied ( Cuddy et al., 2007 ). Cuddy and colleagues (2007) suggested in their model on intergroup affect and behaviors that this ambivalent feeling should elicit discriminating behaviors (e.g., disrespect and neglect; Cuddy et al., 2007 ). Hence, PAD should be higher if older people think their age group is perceived in a paternalistic way and in societies in which they are commonly seen in a patronizing way. Hypothesis 4 (H4): Meta-perceptions of pity felt toward older people predict older adult’s PAD positively at the individual-level (H4a) and societal-level (H4b).

Envious prejudice

For some individuals and societies, older people may be perceived as successfully trying to obtain more control over scarce resources, such as wealth and employment, and are therefore seen as disliked and untrustworthy ( North & Fiske, 2012 ). According to the stereotype content model ( Fiske et al., 2002 ), these feelings of envy should result in discriminatory behaviors (e.g., exclusion and rejection; see Cuddy et al., 2007 ). Hence, older people who think that others perceive them in this way may also report more PAD. Furthermore, older people in societies in which the predominant perception about old age is associated with an envious prejudice should also report more experiences with ageist behavior. Hypothesis 5 (H5): Meta-perceptions of envy toward older people predict older adult’s PAD positively at the individual-level (H5a) and societal-level (H5b).

Contemptuous prejudice

Older people may also be the target of contemptuous prejudice (like other perceived societal free riders such as welfare recipients, see Fiske et al., 2002 ) if they are seen as a burden to society who contribute less than they benefit. Contemptuous prejudice elicits harmful behaviors such as exclusion and demeaning paternalistic behaviors (see Cuddy et al., 2007 ). Hence, older people who think that their age group is met with a contemptuous prejudice and those who live in societies that perceive senior citizens in this way should report more age discrimination. Hypothesis 6 (H6): Meta-perceptions of contempt felt toward older people predict older adult’s PAD positively at the individual-level (H6a) and societal-level (H6b).

Social norms of intolerance of age prejudice

Some individuals are more motivated than others to avoid being prejudiced ( Plant & Devine, 1998 ), and the same may be true of societies (in terms of the majority perspective) whose norms surrounding prejudice and the expression of prejudice can differ. Despite much social psychological research on this construct (see Kunstman, Plant, Zielaskowski, & LaCosse, 2013 ), to the best of our knowledge the motivation to be unprejudiced has not been studied to date as a form of social climate and as a societal-level predictor of social minority’s experience with discrimination.

We propose that prejudice norms are likely to create a social climate in which social minorities feel more or less accepted and that this should also be related to their perceptions of discrimination. A widely shared norm not to express prejudice could be the result of legislative changes protecting the rights of older people, or could stem from widely agreed upon politically correct standards, mandates for proper speech and behavior (e.g., O’Cinneide, 2005 ), as well as cultural values (cf. Vauclair & Fischer, 2011 ), which altogether create an external social pressure in society to value tolerance and avoid prejudice. Hypothesis 7 (H7): The motivation to be age unprejudiced in society—representing a social norm—predicts older adult’s PAD negatively.

Covariates

Van den Heuvel and van Santvoort (2011) conducted a single-level analysis on individual-level predictors of older people’s PAD using ESS data. Because the following variables emerged as significant predictors we included them as covariates in the model: age, gender (1 = male, 2 = female), years of full-time education completed, subjective poverty (1 = living comfortably on present income to 4 = finding it very difficult on present income ), born in country (1 = yes , 2 = no ), perceived seriousness of age discrimination in general in respondents’ country (1 = very serious to 4 = not at all serious ; recoded so that higher scores indicate greater perceived seriousness), life satisfaction (0 = extremely dissatisfied to 10 = extremely satisfied ) and self-rated mental and physical health (1 = very good to 5 = very bad ). 3 Generalized trust was assessed through three questions measuring trust in others (0 = you can’t be too careful to 10 = most people can be trusted ), perceived fairness (0 = most people try to take advantage of me to 10 = most people try to be fair ), and perceived helpfulness (0 = people mostly look out for themselves to 10 = people mostly try to be helpful ). The three items showed satisfactory internal reliability in each country (Cronbach’s alpha ranging from .60 in France to .88 in Cyprus) and were therefore averaged to an overall index of generalized trust.